Genomic instability, one of the hallmarks of cancer, is measured in many forms such as chromosomal instability, microsatellite instability, and instability characterized by increased frequency of base-pair mutations (Bakhoum and Cantley, 2018; Pikor et al., 2013; Negrini et al., 2010). Particularly, chromosomal instability (CIN) is associated with cancer progression, tumor immunity, and inflammation (Pikor et al., 2013; Bach et al., 2019). Recently, CIN has been shown to contribute to other diseases than cancer, including neurodegenerative diseases (Hou et al., 2017; Yurov et al., 2019).
Total Aberration Index (TAI) was proposed by (Baumbusch et al., 2013) to measure the genomic aberrations in serous ovarian cancers. TAI calculates absolute area under the curve for a copy number segment profile generated by piecewise constant fitting (PCF) algorithm. Biologically, TAI can be interpreted as absolute deviation from the normal copy number state averaged over all genomic locations. TAI provides a numerical measure in terms of both prevalence as well as the genomic size of copy number variations in tumors. One of the limitations of TAI is that since it was designed for studying advance stage ovarian tumors and short aberrations found in the early tumors have low impact on TAI. Therefore, TAI should be used to study global scale genomic disorganization that most likely occur in late stage tumors.
tai implemented in CINmetrics takes into account only those sample values that are in aberrant copy number state, i.e. has a mean segment values of less than or equal to −0.2 and greater than or equal to +0.2.
Copy Number Abnormality (CNA) was developed by (Davison et al., 2014) for studying aneuploidy in superficial gastroesophageal adenocarcinoma. An individual CNA is defined as the segment with copy number outside the predefined range of 1.7 to 2.3 with the score of 2 indicating no loss or gain (assuming that the tumor is diploid) as determined by Partek segmentation algorithm. Total CNA for the sample can thus be defined as total number of individual CNAs. CNA can be thought of as a measure of segmental aneuploidy. cna implemented in CINmetrics is similar except we define individual CNA as the segment with copy number less than equal to −0.2 and greater than equal to +0.2 with segment mean of 0 indicating no loss or gain. We chose ±0.2 as a conservative cutoff for TCGA data as described in (Laddha et al., 2014). The users can modify the cutoff by modifying segmentMean parameter.
Counting altered base segments and fraction of the genome altered are modified implementation of the Genome Instability Index (GII) as described in (Chin et al., 2007). The GII was computed in two different ways, both based on calculating common regions of alteration (CRA) and both approaches showed high concordance.
fga is based on identifying common regions of alterations as fraction of the genome altered. Therefore, the fga values are normalized by dividing it by the length of the genome covered. countingBaseSegments on the other hand calculates the common regions of alteration.
All the metrics were computed for TCGASARC, TCGA-OV, TCGA-CRC and log10 scaled in order to compare all the meassurements between them.
tai does not capture this global pattern of difference between normal and tumour samples. tai is best suited for late stage cancers, thus should be used as a measure for studying overall genomic disorganization in individual patients with advanced tumours and not as a measure of genomic instability comparison between normal and tumour samples.
tSNE plot showing the clustering of normal and tumor tissue including HRD34 and the CINmetrics signatures.
There is a correlation between HRD34 and the CINmetric signatures except for tai, which is stated by the cinmetrics publication that is a marker of normal or tumors in an earlier stage. In some types of cancer there seems to be a correlation between low HRD and lower breakpoints and cna, however the small data set limits this conclusions and more statistics are needed.
To mathematically and quantitatively describe these alternations we first locate their genomic positions and measure their ranges. Such algorithms are referred to as segmentation algorithms.
I run CINDex in the masked copy number segments from TCGA-SARC downloaded from GDC data commons. Masked copy number segments repport the segment mean which is the log2(copy-number/ 2). However, CINdex requires the CNV values. Therefore, in order to obtain the copy number value:
$$ CNV = 2*2^{mean value}
$$ Settings: CIN for a treshold gain of 2.25, threshold loss of 1.75, unnormalized (indicated by V.def=3), showing overall (gains and losses) CIN (indicated by V.mode=“sum’).
Summary: * chr1: CIN is significantly higher in UPS HRD high and LMS HRD high. * chr2: CIN is significantly higher in MFS MPNST and UPS in HRD high groups. * chr3: CIN is significantly higher in MFS, LMS and UPS in HRD high groups. * chr4: CIN is significantly higher in MFS, uLMS and UPS and MPNST in HRD high groups. * chr5: CIN is significantly higher in LMS, UPS, uLMS, MFS and MPNST in HRD high groups. * chr6: CIN is significantly higher in MFS, and UPS in HRD high groups. * chr7: CIN is significantly higher in MPNST in HRD high groups. * chr8: CIN is significantly higher in MPNST, MFS, LMS and UPS in HRD high groups. * chr9: CIN is significantly higher in UPS HRD high and LMS HRD high. * chr10: CIN is significantly higher in uLMS in HRD high groups. * chr11: CIN is significantly higher in MFS, LMS, MPNST and UPS in HRD high groups. * chr12: CIN is significantly higher in ddLPS and MPNST in HRD high groups. MFS patient TCGA-DX-AB2S-01A with low HRD shows a very high number of CIN in chromosome 12. * chr13: CIN is significantly higher in UPSand ddLPS in HRD high groups. * chr14: CIN is significantly higher in MPNST in HRD high groups. * chr15: CIN is significantly higher in MPNST in HRD high groups. * chr16: CIN is significantly higher in MPNST, MFS, LMS and UPS in HRD high groups. * chr17: CIN is significantly higher in LMS, ddLPS and UPS in HRD high groups. * chr18: CIN is significantly higher in MFS, MPNST, uLMS and UPS in HRD high groups. * chr19: CIN is significantly higher in UPS in HRD high groups. * chr20: CIN is significantly higher in MFS and UPS in HRD high groups. * chr21: - * chr22: -
In general we can see how HRD high groups show high CIN in individual chromosomes.
There is a High correlation between CINdex and the cna and break points, but there is also certain correlation with fga and base_segments. #### All meassurements
Copyright © 2022 Alicia Pliego. FTP lab. All rights reserved.